The martingale strategy, measured honestly.

Start with 0.1 lots. Lose: trade 0.2. Lose again: 0.4. Any win at any step recovers every previous loss plus the original target profit, and the sequence resets to 0.1. In forex it usually arrives packaged as an EA that keeps adding bigger positions into a losing move (often averaging in every fixed distance), waiting for any pullback to close the whole basket green. On a ranging market — and forex ranges most of the time — the basket exits in profit over and over, and the equity curve looks like a staircase to heaven.

That staircase is the trap. The strategy does not lose often; it loses catastrophically, and rarely enough that a backtest of a few months will happily hide it.

The doubling is exponential and everything it fights is not. With a base lot of 0.1: after 5 straight losses you are carrying 3.2 lots in open risk (32 times the base), after 8 losses, 25.6 lots, after 10, 102.4 lots — the equivalent of a thousand base trades, all in one direction, all underwater. Meanwhile the profit you are chasing at every step stays roughly one base unit. You are risking 1024 units to win 1, and no win-rate makes that arithmetic kind: at a true 50% win rate, a 10-loss streak is a coin-flip event you will meet within a few hundred trades — well inside the life of an account.

The endgame is always the same shape: margin call or stop-out mid-basket, on a trend day, with the account giving back every staircase it ever built. The strategy’s win rate is genuinely high; its expectancy is genuinely negative. Both facts are true at once, and only one of them fits in a marketing screenshot.

The failure mode of martingale lands precisely on the rules that fail most challenges: daily loss and maximum drawdown. A recovery basket growing against a trend produces losses bigger than any single day of profits — one sequence breaches the daily limit by design, not by accident. Firms know the signature too: stacked positions on one side with growing size and no stop is a pattern their behavior reviews flag, whatever the fine print says about EAs. And a martingale EA can pass an evaluation while carrying the bomb: the challenge window is short, the streak may not arrive in time, and the funded account is where it detonates.

If you automate, enforce the cap mechanically, not by memory: Prop Firm Shield blocks execution that would breach the daily loss, which is the only guard that works at 3 a.m. when the fifth recovery step is open and the news hits.

Fixed fractional sizing: every trade risks the same small percentage of equity (the classic 1%), computed from the stop distance with the lot size calculator — losses shrink as equity shrinks, so ruin requires an infinite streak instead of a finite one. Anti-martingale: grow size only after wins and cut it after losses — the same psychology-agnostic math, pointed the right way; it feels worse and survives longer. Cap-and-floor: if you insist on recovery logic, bound it in writing before the session — maximum N steps, total basket risk under a hard ceiling, a daily block — and let software enforce the numbers you agreed with yourself.

All three encode cleanly into the alert pipeline: the JSON carries an explicit lot computed before the alert fires, SignalForge routes it to MetaTrader 5 in milliseconds (the mechanics in the bridge guide, the fields in the JSON examples), and the EA manages it. Rehearse on demo first — the pattern in the paper trading guide — and verify the sizing rule on history before trusting it live: the backtester guide shows how, and backtesting prop firm rules adds the constraints that actually fail accounts. The FOMO guide covers the psychological cousin: recovery trading is usually martingale with feelings instead of math.